Pattern Analysis
Details on individual pattern features.
Last updated
Details on individual pattern features.
Last updated
Once a file has been processed and status reads Success, click the row in the data table to view a preview of the results. For more information on navigating the data table, see Data Management.
To view results from multiple images side by side, click the Workspace icon in the data catalogue row.
The fingerprint overlay shows the diffraction features extracted, the positions of the intensity maxima, and the horizontal spacing in pixels between the specular spot and the 5 nearest non-overlapping diffraction features.
The horizontal spacing, in reciprocal space, can be used to give relative estimates of the lattice constant between two different patterns taken with similar capture settings (electron beam power, sample-screen distance, sample-RHEED gun distance). If a pattern is from a known material (e.g. the substrate) and azimuthal angle, the conversion to real-space units can be applied in the Settings panel using the pixels-to-inverse Angstroms input.
Clicking on the white circle of each diffraction feature shows a subset of the features which are quantified. The primary access point for image results is the API Client, which returns result objects discussed below Programmatic Results
When a video is selected the player will appear below along with the scrub bar. You can jump to a specific frame number, drag and drop the scrubber, or click through on the video player to isolate a specific pattern collected at a moment in time.
To fingerprint the specific frame highlighted in the video player, scroll down to Pattern Analysis -> Saved Frames and click + Save Frame. The extracted patterns from a video will appear below.
Installation instructions for the python AtomCloud API client are listed at API Client.
Programmatic documentation for the methods and classes is available at LINK.
Querying and result fetching are divided into two different steps. Use the search method to query AtomCloud for strings matching file names, physical sample names, or any other associated metadata. This returns a pandas dataframe of matching data items. After optionally filtering this dataframe, use the get method on the id's in the dataframe to fetch the related results.
For single RHEED pattern results, the API client returns lists of RheedImageResult objects.
Create a dataset of the returned result objects RheedImageCollection, which wraps a list of RheedImageResult . This dataset provides an opportunity to add additional metadata to label different data items, which can be used to find correlative relationships between the RHEED features and the external labels.
Details on the convenience methods for working with the featurized RHEED data are available in the programmatic documentation. One key step in working with RHEED datasets is fingerprint alignment. Each diffraction feature is labeled numerically, with the specular spot labeled as feature 0. After querying the collection of RHEED results, use align_fingerprints()to relabel the features such that they match across the dataset. This ensures that downstream pattern comparisons are made between features that are most similarly located relative to the specular spot.
Finally, in order to transform the RHEED features to a tabular format that is friendly for empirical and correlative modeling, use the get_pattern_dataframe() method to generate a pandas dataframe from the underlying networkx graph object representing the RHEED pattern.
